Марія БровінськаAI Eng
2 March 2026, 15:38
2026-03-02
AI, ethics and the Pentagon: why OpenAI and Anthropic quarreled and how Claude overtook ChatGPT
A large-scale conflict has erupted in the United States over the use of artificial intelligence in the defense sector. At the center are OpenAI, Anthropic, their flagship models ChatGPT and Claude, as well as the United States Department of Defense.
The story has already gone beyond the corporate dispute — it is affecting the AI market, federal contracts, and the public debate about the limits of military applications of artificial intelligence.
A large-scale conflict has erupted in the United States over the use of artificial intelligence in the defense sector. At the center are OpenAI, Anthropic, their flagship models ChatGPT and Claude, as well as the United States Department of Defense.
The story has already gone beyond the corporate dispute — it is affecting the AI market, federal contracts, and the public debate about the limits of military applications of artificial intelligence.
What happened
At the end of February, it became known that Anthropic refused to agree to the Pentagon’s demands for a potential expansion of the scenarios for using its Claude model in government systems.
According to American media, the discussion focused on the possible use of AI in mass data analysis tools, use in military systems, and expansion of powers without some of the built-in restrictions.
Anthropic CEO Dario Amodei said the company «cannot in good conscience» agree to remove safeguards if it opens the way to autonomous systems without proper human control.
After that, federal agencies began reviewing their cooperation with Anthropic. Reports appeared in the public space about the suspension of the use of its technologies by some government agencies.
OpenAI enters the game
Against this backdrop, OpenAI has entered into a separate $200 million deal with the United States Department of Defense to use its models in government and likely classified systems.
The company’s CEO Sam Altman publicly emphasized that the agreement includes «red lines», in particular, a ban on use for mass domestic surveillance, a ban on autonomous lethal systems without human intervention, and the presence of technical and legal safeguards.
OpenAI said it would implement additional control mechanisms and audits of the use of models in the public sector.
The market reacted instantly
The scandal quickly went beyond defense policy and turned into a reputational test for the AI market.
After the United States Department of Defense effectively withdrew from cooperation with Anthropic due to its reluctance to loosen safeguards for military use of Claude, the company’s app rose to number one in downloads in the US on Saturday, overtaking ChatGPT.
A movement called «Cancel ChatGPT» has emerged on social media. The quitGPT Instagram account has gained about 10,000 followers in a few days. On Reddit, a post about OpenAI’s contract with the Pentagon garnered over 30,000 upvotes under the title «Cancel and Delete ChatGPT!!!». A chalk inscription appeared outside the Anthropic office in San Francisco: «you give us courage.»
An additional trigger for some users was the mention that OpenAI President Greg Brockman had previously donated $25 million to a super PAC that supported Donald Trump.
The long-term business impact for Anthropic is not yet clear. However, in the short term, the conflict has fueled interest in Claude, provoked a wave of criticism of OpenAI, and launched campaigns calling for the abandonment of ChatGPT.
